DOI: 10.1111/jbi.13383

RESEARCH PAPER

Effects of phylogeny and geography on ecomorphological traits in clades

Anna G. Phillips1,2 | Till Topfer€ 3 | Carsten Rahbek4,5 | Katrin Bohning-Gaese€ 1,2 | Susanne A. Fritz1,2

1Senckenberg Biodiversity and Climate Research Centre (BiK-F), Frankfurt am Abstract Main, Germany Aim: To compare phylogenetic effects with geographic effects across multiple 2Department of Biological Sciences, clades of passerine to understand the roles of evolutionary history and geo- Institute for Ecology, Evolution and Diversity, Goethe University, Frankfurt am graphic patterns on the ecomorphological characteristics of species. Main, Germany Location: Global. 3Zoological Research Museum Alexander Koenig (ZFMK), Section Ornithology, Bonn, Methods: We combine phylogenetic and geographic approaches to investigate and Germany compare their effects on patterns of ecomorphological distinctness, i.e. the relative 4 Center for Macroecology, Evolution and position of species in multidimensional ecomorphological trait space. The trait space Climate, Natural History Museum of Denmark, University of Copenhagen, was based on measurements from preserved specimens, representing ecologically Copenhagen, Denmark relevant morphological adaptations across almost 500 species in eight clades of the 5Department of Life Sciences, Imperial College London, Ascot, UK order Passeriformes. Results: Ecomorphological distinctness increased with phylogenetic distance across Correspondence Anna G. Phillips, Senckenberg Biodiversity species in all clades, whereas there was no significant relationship between geo- and Climate Research Centre (BiK-F), graphic and ecomorphological distinctness in any clade. However, we observed a Frankfurt am Main, Germany. Email: [email protected] significant interaction between phylogenetic and geographic effects on ecomorpho- logical distinctness. Closely related species were ecomorphologically indistinct if in Funding information German Research Foundation, Grant/Award geographic proximity, while at large geographic distances, there was no relationship Number: DFG FR 3246/2-1 between phylogenetic and ecomorphological distinctness.

Editor: Lisa Manne Main conclusions: We conclude that phylogenetic relationships are influential in shaping ecomorphological traits in passerine bird clades, but that this effect depends on the geographic distributions of species. Closely related species were only eco- morphologically similar when geographically close, suggesting a signal of allopatric speciation. Our results imply that studies identifying phylogenetic effects in species’ traits should not focus exclusively on these but instead evaluate the interaction of phylogenetic effects with geographic effects.

KEYWORDS ecomorphological divergence, geographic distribution, morphological adaptations, Passeriformes, phylogenetic signal, trait evolution

1 | INTRODUCTION Recent studies have focused on the evolution of morphological diversity in relation to colonization of new habitats (Jønsson, Les- It is well known that both phylogenetic history and geography shape sard, & Ricklefs, 2015; Ricklefs, 2012), as well as on morphological the evolution and maintenance of morphological diversity in clades. and phylogenetic diversity of species assemblages along

| Journal of Biogeography. 2018;45:2337–2347. wileyonlinelibrary.com/journal/jbi © 2018 John Wiley & Sons Ltd 2337 2338 | PHILLIPS ET AL. environmental gradients (Dehling et al., 2014; Graham & Fine, 2008). geographic effects to interact, as there should be a positive relation- However, there is still a limited understanding of the strength of the ship among ecomorphological and phylogenetic distinctness for geo- effects of phylogenetic relationships and geographic distributions of graphically close but not for geographically widely separated species species on the patterns of morphological diversity of species within (e.g. Cardillo & Warren, 2016). a clade (Simoes~ et al., 2016). The combination of phylogenetic and It is still not clear how ecomorphological divergence (i.e. the evo- geographic effects on morphological diversity within a clade has lution of ecomorphological distinctness) is connected to both the rarely been explicitly tested, and never consistently across multiple phylogenetic history and geographic distribution of species within taxonomic clades (e.g. Harmon, Schulte, Larson, & Losos, 2003; Rick- clades (Simoes~ et al., 2016), and whether one of these effects plays lefs, 2004; Miller, Zanne, & Ricklefs, 2013; but see Lovette, Berming- a more prominent role in shaping ecomorphological diversity than ham, & Ricklefs, 2002). the other. Previous studies suggest that phylogenetic effects rarely We address phylogenetic and geographic effects on the morpho- influence the distribution of ecomorphological traits (Freckleton, Har- logical diversity of clades by studying the ecomorphological distinct- vey, & Pagel, 2002; Harmon et al., 2010; Nyari & Reddy, 2013), yet ness of 491 species in eight avian clades. Ecomorphological some cases have been observed in which signatures of both evolu- approaches are valuable in describing the ecological niche of species tionary and geographic effects can be detected (e.g. Anolis lizards; as morphology often reflects species’ ecology in terms of morpho- Thorpe, Surget-Groba, & Johansson, 2008; Mahler, Revell, Glor, & logical adaptations to the environment (Karr & James, 1975; Ricklefs Losos, 2010). Previous attempts to understand morphological diver- & Miles, 1994; Williams, 1995; Woodward, Winn, & Fish, 2006). We sity have not compared this combination of phylogenetic and geo- define ecomorphologically distinct species as those that have unu- graphic effects on ecomorphological traits consistently across sual or extreme trait combinations in comparison to other species different taxonomic clades (e.g. Harmon et al., 2003; Ricklefs, 2004; within their clade (e.g. Maglianesi, Bluthgen,€ Bohning-Gaese,€ & Sch- but see Lovette et al., 2002). Although inferring evolutionary pro- leuning, 2014). cesses from the ecomorphological diversity that they generate has The distribution of ecomorphological traits within a clade is proven difficult (Warren, Cardillo, Rosauer, & Bolnick, 2014), a com- affected by the shared ancestry of closely related species that are parative approach across multiple clades could test the consistency often more morphologically similar than distantly related species of these patterns. (Losos, 2008). Hence, some phylogenetic effects on ecomorphologi- We combine ecological and evolutionary approaches by testing cal distinctness would be expected, signified by an increase in eco- how ecomorphological diversity of species within eight different morphological distinctness with greater phylogenetic distances avian clades is related to both the phylogenetic history and the geo- between species, i.e. greater phylogenetic distinctness. As a conse- graphic distributions of the species. If phylogenetic history leaves a quence, a particularly steep increase in ecomorphological distinctness signal on ecomorphological diversity, we expect that phylogenetically with phylogenetic distinctness is expected under phylogenetic niche distinct species, i.e. more distantly related species are also ecomor- conservatism with particularly high morphological similarity in closely phologically distinct. If competition leaves a signal on ecomorpholog- related species (e.g. Ackerly, 2009). By contrast, small increases in ical patterns, we expect that geographic distinctness (i.e. greater ecomorphological distinctness with phylogenetic distinctness or no geographic distances between species distributions) is negatively relationship between the two would be expected under a process, related to ecomorphological distinctness (i.e. that more closely among others, of convergent evolution (e.g. Harmon, Kolbe, Che- located species are more distant in ecomorphological space). Alterna- verud, & Losos, 2005). In any case, the presence of some phyloge- tively, if allopatric speciation leaves a signal on ecomorphological dis- netic effect is expected even under simple evolutionary models such tinctness, we expect an interaction effect of geographic with as Brownian motion, i.e. trait evolution resembling a random walk phylogenetic distinctness. We address this question across eight through time (Losos, 2008). avian clades that differ in phylogenetic age and geographic distribu- The geographic distribution of species within a clade may also tion, to assess whether patterns are general among different taxo- influence ecomorphologial distinctness due to processes such as nomic groups. We compare radiations of similar species richness interspecific competition or allopatric speciation. If geographically within one order, facilitating the comparison of patterns across the overlapping species compete with each other, and if the degree of selected clades. competition is correlated with similarity in ecomorphological traits, co-occurring (sympatric) species would be expected to be ecomor- 2 | MATERIALS AND METHODS phologically more distant than geographically nonoverlapping (al- lopatric) species, to avoid interspecific competition (e.g. Davies, 2.1 | Data preparation Meiri, Barraclough, & Gittleman, 2007; Rodrıguez-Girones & Santa- marıa, 2007). Alternatively, if allopatric speciation leaves a geo- We selected eight monophyletic clades across the order of passerine graphic signal on the ecomorphological diversity of species, species birds (Aves: Passeriformes) to test the influence of phylogenetic and that are close in geographic space (allopatric or in secondary sympa- geographic effects on the ecomorphological distinctness across spe- try) would be expected to be ecomorphologically similar and phylo- cies among and within the clades. have a relatively uni- genetically close. Thereby one would expect phylogenetic and form morphology, which enables a better comparison of similar PHILLIPS ET AL. | 2339 structures across multiple families. Clades of different ages (between and PC2 as these axes represented 94.2% of the explained variance approximately nine and 24 million years) have been selected from in our data (Table S2.2 in Appendix S2). The trait space included all different parts of the passerine tree, and constitute families or eight clades to ensure comparability, but analyses of distinctness monophyletic subclades of families (Cardinalidae, Parulidae: Seto- were conducted within each clade separately. To directly determine phaga-Myiothlypis clade, Muscicapidae: Oenanthe-Monticola clade, the effect of phylogenetic distinctness of a species on its position in Turdidae: Turdus, Hirundinidae, Vireonidae, Corvidae: genus ecomorphological trait space, we obtained dated phylogenies from Corvus, and Tyrannidae: Xolmiini clade; see Appendix S1 in Support- birdtree.org (Jetz, Thomas, Joy, Hartmann, & Mooers, 2012) and cre- ing Information for species lists and Appendix S2 for genera). These ated consensus trees for each clade (1,000 trees per clade, 25% were chosen based on the following criteria: they must (a) have burn-in removed, 95% maximum clade credibility) using TreeAnnota- approximately the same number of species (see Appendix S1), (b) tor, enabled in BEAST 1.8 (Drummond, Suchard, Xie, & Rambaut, have high phylogenetic resolution, (c) represent a variety of feeding 2012). Geographic data of breeding ranges were obtained for all spe- strategies and dietary guilds, and (d) show a considerable degree of cies of all clades (Holt et al., 2013) to determine the effect of geo- morphological trait diversity within the clade. The clades have vary- graphic distances and overlap among species. ing geographic distributions (Xolmiini, Vireonidae, Setophaga-Myioth- lypis, and Cardinalidae: Americas; Oenanthe-Monticola: Asia, Africa, 2.2 | Distinctness quantification and Europe; Hirundinidae: Worldwide; Corvus: Worldwide except South America; Turdus: Worldwide except Australia). Species names To determine the phylogenetic and geographic effects on ecomor- follow IOC v. 5.01 (Gill & Donsker, 2015; see Table S1.1 phological distinctness, we calculated average distinctness values for in Appendix S1 and supplementary methods in Appendix S2). each species across all species pairs within the clade in ecomorpho- We quantified the ecomorphological traits of 491 species in logical trait space (mean ecomorphological distinctness, MED), on these clades using nine morphological trait measurements (of the the phylogeny (mean phylogenetic distinctness, MPD), and in geo- beak, wings, tail, and tarsi) of preserved specimens. The selected graphic space (mean geographic distinctness, MGD). We also calcu- measurements are closely related to specific ecological niche dimen- lated the distances between nearest neighbour pairs in sions such as diet and foraging behaviour (e.g. Grant & Grant, 2006; ecomorphological trait space (nearest neighbour ecomorphological Jønsson et al., 2012), aerial movement and dispersal distance (e.g. distance, NNED) and the phylogenetic and geographic distances Calmaestra & Moreno, 2000; Dawideit, Phillimore, Laube, Leisler, & between these pairs (nearest neighbour phylogenetic distance, Bohning-Gaese,€ 2009), as well as bipedal locomotion (e.g. Fitzpatrick, NNPD, and nearest neighbour geographic distance, NNGD). 1985), and can thus be used to reflect ecomorphological relation- Ecomorphological distinctness for each species was measured as ships (e.g. Ricklefs, 2012; Winkler & Leisler, 1992). We measured the mean pairwise ecomorphological distance (MED) to all other spe- morphological traits of 2,465 preserved specimens belonging to 491 cies within the clade in trait space, where a greater value indicates species of the total 526 species described in the eight clades from greater distinctness (Laliberte & Legendre, 2010). This measure was four museum collections (Table S1.1 in Appendix S1). We only took chosen over morphological “originality” of each species (distance to trait data from adult individuals and aimed to measure two females the centroid) as we wanted to determine the ecomorphological posi- and two males of each species. The methodology largely followed tion of each species relative to all other species of the clade rather Eck et al. (2011) except for bill dimensions (see supplementary meth- than to a mean clade value. Ecomorphological distances between ods in Appendix S2 for full description). We additionally measured nearest neighbours in trait space (NNED) were calculated by extract- any available morphologically distinct subspecies to account for ing the shortest distance to a neighbouring species within the same intraspecific variation. In many cases, we therefore obtained data clade from the pairwise ecomorphological distance matrix for each from more than four individuals per species (average number of indi- species (see Dehling et al., 2014). viduals per species = 5.03, min = 1, max = 29). To account for varia- Pairwise phylogenetic distances were calculated between all pos- tion in body sizes across all clades, the body mass of each species sible species pairs for each clade. Using these distances, we calcu- was included as the tenth variable in analyses, obtained from a pub- lated the mean phylogenetic distance (MPD) between species pairs, lic database (Wilman et al., 2014). where a greater mean distance indicates greater distinctness (Webb, Prior to analysis, measured values for each trait were averaged Ackerly, McPeek, & Donoghue, 2002). We also calculated the phylo- across all specimens in each species irrespective of sex. We checked genetic distance between each nearest ecomorphological neighbour that within-species trait variation was lower than the variation pair (NNPD). We consider these methods more relevant to our between species (data not shown). Where available, trait data for objective than the phylomorphospace approach (Revell, 2012; Sid- morphologically distinct subspecies were included unweighted in the lauskas, 2008), as we wanted to quantify phylogenetic signal rather overall species average. Species averages were log-transformed, then than control for it. standardised to a mean of 0 and a standard deviation of 1. To The geographic range data were used to address the degree of reduce dimensionality, species ecomorphological trait averages and range overlap between species within a clade (Cardillo & Warren, body mass were subjected to a combined principal components anal- 2016). Across all species pairs, the mean of pairwise geographic dis- ysis (PCA). Ecomorphological trait space was characterised using PC1 tances between each of the grid cells of one species’ breeding range 2340 | PHILLIPS ET AL. to all grid cells of another species’ breeding range was calculated to Furthermore, we used two methods to test the relative strengths obtain a value of mean geographic distinctness (MGD) for each spe- of phylogenetic and geographic effects on ecomorphological distinct- cies, where a greater mean distance indicates greater distinctness. ness. Firstly, we constructed a mixed-effects model with both phylo- MGD values were log-transformed, as we observed a large variation genetic and geographic distinctness as fixed effects across all species in the scale of MGD within and between the clades. Using the pair- in all clades. We constructed an equivalent model for nearest eco- wise distance matrices, we additionally extracted the geographic dis- morphological neighbour distances with both the phylogenetic and tance between nearest ecomorphological neighbour pairs (NNGD), geographic distances of nearest ecomorphological neighbours as which were also log-transformed. fixed effects across all clades. Both models additionally assessed whether the phylogenetic variable significantly differed in its rela- tionship of the ecomorphological distinctness with the geographic 2.3 | Statistical analyses variable by including an interaction of MPD with MGD, and NNPD We tested five different methods to determine the phylogenetic with NNGD. Secondly, we inferred the relative contributions of phy- effect on the patterns of ecomorphological distinctness in trait logenetic and geographic distinctness to the ecomorphological dis- space: (a) Mixed-effects models between ecomorphological and phy- tinctness of species within each clade using the method described logenetic distinctness values across all species in all clades; (b) by Freckleton and Jetz (2009). While our mixed-effects model simply regressions of the same relationship within each clade; (c) mixed- looks for a relationship between observed ecomorphological and effects models equivalent to (a) across all nearest neighbour dis- phylogenetic or geographic patterns, Freckleton and Jetz’s method tances in all clades, (d) regressions of the same relationship across does not estimate interaction effects. Instead it uses an a Brownian nearest neighbour distances within clades; and (v) phylogenetic sig- motion model of trait evolution as the null model (k-statistic; Pagel, nal of MED within clades (k; Pagel, 1999). These methods test for a 1999; Freckleton et al., 2002) to estimate the phylogenetic and geo- phylogenetic effect against different null models: (a) and (c) no phy- graphic effects in each clade by directly incorporating a variance- logenetic effect (slope estimate of zero) in the mixed-effects models covariance matrix for the phylogenetic relationships and a pairwise (for MED and NNED); (b) and (d) a random-shuffle null model (i.e. distance matrix for geographic ranges of species. The values of the random shuffling of species’ identities across the tips of the phy- geographic distance matrix were first log-transformed, and then logeny) to account for the topology of the phylogenetic tree in standardised to range between 0 and 1. This method estimates three regression analyses within clades (for MED and NNED); (v) two null parameters, which represent the relative contributions of phyloge- models, the random-shuffle model and a Brownian motion process netic effects (k’), geographic effects (φ) as well as unexplained effects of trait evolution (random walk of evolution), in phylogenetic signal (c) that are independent of phylogenetic and geographic effects. analyses (for MED only). Mixed-effects models were fitted across all species in all eight clades, where MED was the response variable and MPD the fixed 3 | RESULTS variable. We allowed random slopes and intercepts for each clade, to model clade variation in relationships between ecomorphological and The constructed ecomorphological trait space represents the position phylogenetic distinctness. We ran equivalent models between NNED of each species relative to all others in the clade based on their eco- and NNPD values to further determine the phylogenetic effect on morphological traits, and the arrangement of all eight clades in this ecomorphological distances between nearest neighbours. To assess trait space relative to each other (Figure 1). Factor loadings indicated whether clades significantly differed in their relationships of ecomor- that all variables were approximately equally important for PC1; this phological to phylogenetic distinctness, the mixed-effects models axis therefore represented a proxy of overall body size (representing across all clades (for MED and NNED) were also fitted with clade as 82.6% of the variation, Table S2.2, S2.3 in Appendix S2). PC2 repre- a fixed effect and the interaction of clade with the phylogenetic pre- sented wing shape and tarsus length (explaining 11.6% of the varia- dictor variable. In addition, regression models within clades tested tion, Figure 1, Table S2.3). The PC scores on these two axes were the phylogenetic effect on ecomorphological distinctness in each used to characterize the ecomorphological trait space of the species clade separately against the random-shuffle null model (for MED and (see Methods; Figure 1). NNED). As a more conventional measure of the phylogenetic pattern in trait values across species, the phylogenetic signal metric k was 3.1 | Phylogenetic and geographic effects on calculated to distinguish the observed phylogenetic pattern from ecomorphological distinctness both a random phylogenetic distribution and from the distribution expected under a Brownian motion process of trait evolution. The mixed-effects model across all species in all clades showed a sig- To determine the effect of geographic distinctness on the patterns nificant positive relationship between ecomorphological and phyloge- of ecomorphological distinctness, we used the equivalent methods (a)- netic distinctness, meaning ecomorphological distinctness increased (d) for geographic distributions of species with the same null models. with phylogenetic distance between species (Table 1, Figure 2a). We tested for significant differences between clades in these models as Clade never significantly affected the other fixed effects when described above for the phylogenetic mixed-effects models. tested as a fixed effect with interaction term (with MPD, MGD, PHILLIPS ET AL. | 2341

rounded wings Cardinalidae long tarsi Corvus Hirundinidae ● ● Oenanthe-Monticola 1 ● ● ● ● Setophaga-Myiothlypis ● ●●● ● Turdus ● ●●●● ● ●●●●●● Xolmiini ●● ● ●● ● ●●●● ● Vireonidae ●● ●●● ● ● ●●●●●● ● 0 ●●● ● ● ●●● PC2 −2 −1

pointed wings short tarsi −4 −2 0 246810 small body large body size size PC1

FIGURE 1 The ecomorphological distribution of 491 species in eight monophyletic clades (Cardinalidae, Corvus, Hirundinidae, Oenanthe- Monticola clade, Setophaga-Myiothlypis clade, Turdus, Xolmiini clade and Vireonidae), indicated by coloured symbols. The PCA was run across species averages of ten traits: nine ecomorphological measurements (wing length and pointedness (measured as Kipp’s distance); bill length, width and height; tail length; tarsus length, as well as sagittal and distal tarsal diameters) and body mass. Higher scores for PC1 were associated with larger body size, while lower scores were associated with smaller body size. PC2 represented wing shape and tarsus length, where high scores for PC2 were associated with rounded wings (i.e. smaller Kipp’s distance) and longer tarsi, while lower scores indicated pointed wings (i.e. higher Kipp’s distance) and shorter tarsi

NNPD or NNGD; Table S2.4 in Appendix S2), therefore we con- Figure 2b). Clade-level regressions of ecomorphological and geo- trolled for clade as a random effect in the main models presented graphic distinctness showed an apparently greater variation in slopes here (Table 1). Regressions within clades showed an overall positive among clades than we observed between ecomorphological and phy- trend between ecomorphological and phylogenetic distinctness, but logenetic distinctness. None of these slopes significantly differed significance compared to the random-shuffle null model varied from expectations under the random-shuffle null model (Table 2, Fig- across clades (Table 2, Figure 2a). Ecomorphologically distinct spe- ure 2b). The explanatory power in these models was weak, with cies were significantly distant on the phylogeny in Oenanthe-Monti- most clades having R2 < 0.1 values. Similarly, the distance between cola and allies (Muscicapidae), Xolmiini (Tyrannidae) and Vireonidae. nearest ecomorphological neighbours could not be explained by geo- The explanatory power in these models was weak, with most clades graphic distances between these species pairs within or across having R2 < 0.1 values. We estimated Pagel’s k of ecomorphological clades, supporting our other findings (mixed-effects model across distinctness as a direct measure for phylogenetic signal, and found clades in Table 1, within-clade regressions in Table 2). some variation of phylogenetic signal among the clades. Pagel’s k was significantly higher than expected from the random-shuffle null 3.2 | Combined effects of phylogeny and model in all but two clades, but only significantly different from the geography Brownian motion null model in four clades (Table 2). In contrast to ecomorphological distinctness, the ecomorphologi- The mixed-effects model investigating the combined effects of phy- cal distance between nearest ecomorphological neighbours was not logenetic and geographic distinctness on ecomorphological distinct- explained by their phylogenetic distances across clades (mixed- ness across clades showed positive phylogenetic and geographic effects model, Table 1), though we found a significant negative rela- effects on ecomorphological distinctness, and also a significant nega- tionship of NNED and NNPD between species pairs in one of the tive interaction effect between these fixed effects (Table 1). The within-clade models (Xolmiini, Table 2). This implies that nearest geographic effect on ecomorphological distinctness was only signifi- ecomorphological neighbours in Xolmiini, which are separated by cant in the interaction with phylogenetic distinctness, where the small distances in trait space, are more distantly related on the phy- relationship between ecomorphological and phylogenetic distinctness logeny than expected under the random-shuffle null model. was positive only at low values of geographic distinctness, but We compared the ecomorphological distinctness of species nonexistent at high values (Figure 3). In the combined mixed-effects across all clades with their geographic distinctness in a mixed-effects model for NNED, neither phylogenetic nor geographic distances model to infer the geographic effects on species’ distributions in between nearest ecomorphological neighbour pairs significantly influ- ecomorphological space and found a negative, but nonsignificant enced their distance in ecomorphological trait space, nor did we relationship (Table 1, Figure 2b). Furthermore, we found no signifi- observe a significant interaction effect between NNPD and NNGD cant differences in the relationships among clades (Table S2.4, (Table 1). 2342 | PHILLIPS ET AL.

TABLE 1 Results of mixed-effects models across all species of all clades testing the relationships of mean ecomorphological distinctness (MED) against mean phylogenetic distinctness (MPD) and mean geographic distinctness (MGD) separately and in combination, and the same models for nearest neighbour ecomorphological distances (NNED) against their phylogenetic (NNPD) and geographic distances (NNGD). Combined models include a test for an interaction between the phylogenetic and geographic variables (indicated by *). Number of species analysed in each model are indicated by n. Mixed-effects models detail the overall slope estimate, its t-value, degrees of freedom (d.f.), and its significance (p); the estimated variation in slopes among clades (var, random effect); and goodness of fit of the model to the data (conditional R2)

n Estimate t d.f. p var R2 MPD 456 0.045 4.006 5.899 0.007a <0.001 0.676 MGD 453 À0.141 À0.640 9.299 0.537 0.003 0.476 NNPD 456 À0.001 À1.564 2.944 0.218 <0.001 0.115 NNGD 453 <À0.001 À0.192 302.100 0.848 <0.001 0.098 Combined effects of MPD 451 0.443 2.997 35.170 0.005a <0.001 0.580 MGD 1.076 2.640 31.270 0.013a 0.076 MPD*MGD À0.047 À2.752 34.510 0.009a Combined effects of NNPD 451 À0.003 À1.479 6.860 0.144 <0.001 0.113 NNGD À0.003 À0.725 32.630 0.469 <0.001 NNPD*NNGD <0.001 0.813 31.730 0.417 ap < 0.05.

The results from partitioning the variance in ecomorphological described effects are not entirely consistent down to the level of distinctness across species into phylogenetic and geographic effects species pairs that are closest in ecomorphological trait space. corroborated these results, but also showed a general pattern of While phylogenetic effects were more influential than geographic decreasing phylogenetic effects with decreasing taxonomic level (i.e. effects in shaping the traits of species across clades, these effects from family to subclade to genus level; Figure 4). In most clades, depended on the geographic distribution of species. In close geo- ecomorphological distinctness seemed to be heavily influenced by graphic proximity, closely related species were also closer in ecomor- phylogenetic effects (k’), while there were substantial unknown phological trait space (i.e. there was a significant positive relationship effects (c), which were independent of phylogenetic and geographic between ecomorphological distinctness and phylogenetic distinct- effects (Table S2.5 in Appendix S2). This relationship was not ness). In contrast, when species were not in geographic proximity, observed in Turdus, in which ecomorphological distinctness was this relationship was not observed. Close geographic proximity in explained completely by independent effects (c). We observed no our case does not distinguish whether species ranges are completely, direct geographic effect in any of our clades (Figure 4). partly or not overlapping, so geographically close species could be sympatric (i.e. coexisting in the same geographic area) or allopatric (i.e. occurring in separate, nonoverlapping geographic areas). If sym- 4 | DISCUSSION patric species compete, a negative relationship between geographic distance and ecomorphological distinctness is expected (Dayan & The objective of this study was to test the effects of phylogenetic Simberloff, 2005). We found instead that geographic distance modi- and geographic distance on distances in ecomorphological trait space fied phylogenetic effects on ecomorphological distinctness, which of species within and across eight clades of Passerines. Effects of could suggest that these patterns have been shaped by a process of phylogenetic and geographic distinctness were not independent of allopatric speciation. Under such a scenario, comparatively recently each other, and the significance of the geographic effect on ecomor- split species would still be ecomorphologically relatively similar due phological diversity was influenced by the phylogenetic distinctness to a lack of competitive effects on ecomorphological divergence of species. When testing phylogenetic and geographic effects sepa- (Pigot & Tobias, 2013; Price, 2008). rately across species in the eight individual clades we observed con- Similar to findings in another single clade of birds (Tobias et al., sistent phylogenetic effects on ecomorphological distinctness, but no 2014), we found that when species were not in geographic proximity direct geographic effects. These patterns across clades were fairly (i.e. had clearly separated ranges) both distant and close relatives homogeneous despite having selected clades of different phyloge- could be ecomorphologically similar. This pattern could be driven by netic ages and geographic distribution patterns, with no significant a mix of processes, including ecological and geographic opportunity. differences among clades. Significant effects were only observed for The geographic separation of species may allow these to diversify in MED, not nearest neighbour ecomorphological distance, so the traits after they enter a new habitat with no ecological limits to PHILLIPS ET AL. | 2343

(a) Cardinalidae Corvus Hirundinidae Oenanthe-Monticola Setophaga-Myiothlypis Turdus Xolmiini Vireonidae Ecomorphological distinctness (MED) 0.5 1.0 1.5 2.0 2.5 3.0 15 20 25 30 35 40 45 Phylogenetic distinctness (MPD)

(b) Ecomorphological distinctness (MED) 0.5 1.0 1.5 2.0 2.5 3.0 2500 3500 5000 7500 Geographic distinctness (MGD)

FIGURE 2 Mixed-effects models across all species with random slopes and intercepts for each clade, for ecomorphological distinctness (MED, the mean pairwise distance of a species to all other species in the same clade measured in the trait space shown in Figure 1) as response variable, and for the fixed effects (a) phylogenetic distinctness (MPD, the mean pairwise phylogenetic distance to species in the same clade measured in million years) and (b) geographic distinctness (MGD, the mean pairwise geographic distance to species in the same clade measured in km and log-transformed). Both models incorporated clade allocation as the random effect. Each clade is shown in a different colour, where each point represents one species. Observed slopes for each clade are from the model across clades; they are shown as solid if the correlations were significant in the linear regressions of the individual clade, and as dashed if not. Additionally we show 95% confidence intervals derived from a null model for each clade (shaded polygons; random shuffling of phylogenetic (a) and geographic (b) distinctness values across the entire clade)

diversity (ecological opportunity; Wiens, 2011), so that they may speciation rates and diversification patterns have frequently been become dissimilar to their geographically remote relatives; or they considered (Barrera-Guzman, Mila, Sanchez-Gonzalez, & Navarro- may maintain their traits as they would be released from congeneric Sig€uenza, 2012; Graham, Ron, Santos, Schneider, & Moritz, 2004; competition (geographic opportunity; Simoes~ et al., 2016). These Mittelbach & Schemske, 2015; Near & Benard, 2004), few studies patterns have been observed in the Caribbean Anoles, where dis- have synthesized these approaches in relation to morphological vari- tantly related species living in allopatry have developed remarkably ation (Harmon, Melville, Larson, & Losos, 2008; Velasco et al., 2016). similar traits (Harmon et al., 2005; Mahler et al., 2010). These spe- Of the studies that have considered the influence of both historical cies show a decrease in adaptive differentiation among islands, fol- and spatial data, some also find an influence of geography on mor- lowing initial increases in disparity through the availability of phological evolution (e.g. Harmon et al., 2008). Therefore, especially ecological opportunities. when studying groups of species, their geographic distributions The nonindependence of phylogenetic and geographic effects should be controlled for, particularly for sympatric species. affecting the distribution of species in trait space shows that the When testing the phylogenetic effects without consideration of phylogenetic pattern in morphological diversity cannot be inter- the influence of geography on these patterns, we found consistent preted accurately without also considering the geographic proximity effects on ecomorphological distinctness across all species in all of species. The importance of jointly investigating phylogenetic and clades, but the strength of these effects within clades appeared to be geographic influences on species’ traits is also emphasized by Freck- quite variable. Due to the influence of geography on the phylogenetic leton and Jetz (2009), who show that traits often exhibit both phylo- effect on ecomorphological distinctness, it is difficult to interpret genetic and spatial structures. While their combined influences on these results without also considering the contribution of geographic 2344 | PHILLIPS ET AL.

3.2 2 R

7500

5000 0.006 0.156 0.017 0.0080.001 0.154 0.427 0.027 0.001 Slope p.null À À À stance (MPD) and mean 2 Geographic distinctness (MGD) Geographic R

3500 Ecomorphological distinctness (MED) on null model. Number of species ) and geographic distances (NNGD). This ) 2 R 2500 0.5 15 20 25 30 35 40 45 Phylogenetic distinctness (MPD) 1.345 0.148 0.025 0.003 0.427 0.002 0.196 0.420 0.004 0.005 0.228 0.010 1.578 0.150 0.026 Slope p.null À MGD NNGD À À FIGURE 3 Trend surface plot showing the interacting effects of phylogenetic and geographic distinctness on ecomorphological Geographic effects n distinctness as estimated in the combined model across all species in the eight clades. Circles depict the position of the observed

2 species data, and circle size corresponds to the level of observed compared to two null models: no phylogenetic signal, i.e. random shuffling of R

k ecomorphological distinctness. The colour surface shows the fitted relationship of ecomorphological distinctness (MED) with both phylogenetic (MPD) and geographic distinctness (MGD) across 451 species 1). For regression models, each section details the estimated slope; significance of = k phylogenetic 0.004 0.036* 0.082 43 0.003 0.149 0.015 68 0.812 0.074 0.033 0.001 0.260 0.007 65 0.001 0.220 0.008 79 0.441 0.395 0.001 0.001 0.332 0.002 0.006 0.083 0.057 42 unexplained spatial Slope p.null À À À À À 2 ) is shown with significance of R k % effect 0.001* 0.435 0.000 0.347 0.002 47 1.743 0.059 0.073 0.007 0.127 0.026 0.001* 0.185 < < 0.0 0.2 0.4 0.6 0.8 1.0 Turdus Corvus Xolmiini Vireonidae Cardinalidae Hirundinidae Oenanthe-Monticola Setophaga-Myiothlypis 1) Slope p.null = k 0.001* 0.003 0.339 0.001 0.001* 0.075 FIGURE 4 Partitioned contributions of phylogenetic, spatial and < < unexplained independent effects on the variance in ecomorphological distinctness of species within each clade. . The phylogenetic signal of each group ( 0) p( n

= Variance was partitioned into phylogenetic (k’, dark grey), spatial (φ) k 0.001* 0.078 0.067 0.999 0.001* 0.001* 0.549 0.076 0.180 0.019 0.001 0.170 0.018 45 0.517 0.347 0.004 0.001* 0.247 0.034 0.077 0.033 and unexplained independent (c, light grey) components. The p( < > < < < results were ordered by the strength of the phylogenetic effect, and showed no spatial effect on ecomorphological distinctness in any of our clades Phylogenetic signal MPD NNPD k 0); and phylogenetic signal in accordance with Brownian motion process p(

= distribution. Although we observed consistently positive phyloge- Phylogenetic effects n 6967 0.618 0.000 0.015* 0.011* 0.033 0.125 0.012 0.001 0.361 0.003 64 0.166 0.344 0.002 0.001 0.378 0.001 62 0.781 41 0.375 0.359 0.069 0.037 0.106 0.039 k netic effects across clades overall, this variation in strength within Results of separate linear regression models for each clade of mean pairwise ecomorphological distance (MED) against mean pairwise phylogenetic di clades could suggest that different processes influence patterns of ecomorphological distinctness to different degrees in each clade.

0.05. Previous studies have distinguished between different phylogenetic < XolmiiniVireonidae 42 49 0.840 0.916 0.005* 0.002* 0.101 0.032* 0.087 Setophaga Turdus Cardinalidae 48 0.906 Oenanthe Corvus Hirundinidae 78 0.938 p * pairwise geographic distance (MGD), andshows the test same results models against for theanalysed nearest random-shuffle in neighbour null each ecomorphological model, clade distances and arespecies (NNED) the indicated on against phylogenetic by tree their signal p( phylogenetic in (NNPD MED tested against the random-shuffle and the Brownian moti TABLE 2 phylogenetic/geographic isolation compared to the random-shuffle null model (p.null); and goodness of fit of the regression model to the data ( patterns in morphological diversity, for example citing processes PHILLIPS ET AL. | 2345 such as phylogenetic niche conservatism or weaker evolutionary have influenced the ecomorphological divergence of species (Maire, effects as having driven these patterns (Bravo, Remsen, & Brumfield, Grenouillet, Brosse, & Villeger, 2015), e.g. selective pressures associ- 2014; Hawkins, Diniz-Filho, Jaramillo, & Soeller, 2007; Miller et al., ated with their habitat or climatic niche, competition with sympatric 2013). Other studies investigating species’ occupation of niche space species of other clades, or other behavioural and morphological have come to similar conclusions, observing strong phylogenetic sig- aspects (cf. Nyari & Reddy, 2013; Velasco et al., 2016). These were nal but no independent influential geographic signal (e.g. Miller et al., beyond the scope of this study but would be important to investi- 2013; Verbruggen et al., 2009). Our finding of a significant interac- gate in future analyses. As phylogenetic signal is known to be tion with geographic effects leads us to conclude that such results weaker in ecological traits than in morphological traits (Blomberg, need to be evaluated with the geographic influences on phylogenetic Garland, & Ives, 2003; Bohning-Gaese€ & Oberrath, 1999; Freckleton patterns in morphological diversity taken into account (Pigot & et al., 2002), choosing ecomorphological traits in our comparative Tobias, 2013). Due to lower statistical power we could not test the study across clades retained this phylogenetic signal of morphologi- interaction between phylogenetic and geographic effects in the indi- cal traits, while still accounting for temporal or geographic variation vidual clades. Instead we tested individual null models within all of the ecological niche position of species. clades. Mixed-effects models, regressions, phylogenetic signal, and Our study adds to scarce literature comparing patterns of trait variance partitioning each used different approaches to calculate the distribution within and across clades (e.g. Harmon et al., 2010; Love- phylogenetic effect on ecomorphological distinctness. However, the tte et al., 2002), but contrasts to previous findings in that it suggests different approaches across clades agree on strong phylogenetic and a homogeneous ecomorphological trait distribution pattern across weak or no geographic effects, supporting our interpretations above clades despite different ages and geographic distributions of the despite observed variation in within-clade results, which may also be eight clades. The ability of the models to detect a great amount of caused simply by the lower sample size within clades. between-clade variation is reduced due to our limited number of Geographic distinctness of species within clades had no direct clades. Nevertheless, as previous studies have often used studied a effect on patterns of ecomorphological distinctness alone, but influ- single clade, our analysis of eight independent clades therefore pro- enced the relationship between phylogenetic and ecomorphological dis- vides a broader inspection of the patterns of ecomorphological tinctness. Our results suggest that the geographic signal on its own is divergence. In previous studies, the rates of morphological evolution either completely absent or has not been preserved within these clades. and speciation have both been shown to depend on clade age, No geographic signal of competition would be preserved if allopatric where younger clades often show higher rates of morphological evo- speciation with later sympatry is observed for ecomorphologically simi- lution than older clades (Harmon et al., 2010; Ricklefs, 2004). Thus, lar species (Cardillo & Warren, 2016; but see Pigot & Tobias, 2013). We we expected to see variable ecomorphological distinctness patterns therefore also lack evidence for direct competition effects in these across clades with differing phylogenetic and geographic scales. This clades, though competition might have played a role in secondary con- variation was not observed, despite selecting clades of similar spe- tact among closely related species. The lack of geographic signal may cies richness. Instead, clades in which relationships between ecomor- also be due to our measure of geographic distinctness, which is mea- phological and phylogenetic distinctness were statistically significant sured only at a broad geographic scale (i.e. the degree to which extent- span different phylogenetic and geographic scales. of-occurrence range maps in general overlap). Variations in geographic scale have previously been suggested to influence patterns of species distributions and diversity, and may therefore also influence patterns of 5 | CONCLUSIONS morphological diversity within clades (Cardillo & Warren, 2016; Graham & Fine, 2008). We observed no geographic signal at the relatively rough In combining the patterns of morphological adaptations, phylogenetic spatial grain (100 9 100 km grid cells) of the geographic ranges of our relationships and geographic distributions of species, we were able species, though some signal may be observed at smaller scales of habi- to compare the roles of evolutionary history and geographic distribu- tat partitioning, which is beyond the spatial scale captured in our geo- tion of species in shaping trait diversity in clades. When comparing graphic distinctness measure. It is well known that phylogenetically phylogenetic with geographic effects, the former appeared more close, ecomorphologically similar and competing bird species separate important in determining distances between species in ecomorpho- themselves across space much more finely, for example across different logical trait space. Our measure of geographic proximity did not habitat types in the same region (e.g. Sylvia warblers in Mediterranean appear to directly influence the ecomorphological distinctness of regions; Laube, Graham, & Bohning-Gaese,€ 2013 and references species in our clades. However, the geographic distance among spe- within), or across different canopy layers in the same forest (e.g. Seto- cies modified the relationship between ecomorphological and phylo- phaga warblers; MacArthur, 1958). Across clades, however, the interac- genetic distinctness, which we interpret as a potential signal of tion between geographic and phylogenetic effects implies that allopatric speciation across clades; therefore, taking geography into geographic effects are not independent from phylogenetic effects. account should become standard in future studies of trait diver- Our measure of ecomorphological divergence in trait space was gence. We conclude that applying ecological community methods to based on ten traits, and was further reduced to the two main axes monophyletic clades does not sufficiently capture the effects of of the PCA. This may mask divergence in other traits, which may some processes on trait divergence within clades, such as 2346 | PHILLIPS ET AL. interspecific competition. However, these methods can provide Fitzpatrick, J. W. (1985). Form, foraging behavior, and adaptive radiation – interesting insights into the interactions of trait evolution and geo- in the Tyrannidae. Ornithological Monographs, 36, 447 470. Freckleton, R. P., Harvey, P. H., & Pagel, M. (2002). Phylogenetic analysis graphic distributions of species and phylogenetic lineages. and comparative data: A test and review of evidence. The American Naturalist, 160, 712–726. Freckleton, R. P., & Jetz, W. (2009). Space versus phylogeny: Disentan- ACKNOWLEDGEMENTS gling phylogenetic and spatial signals in comparative data. Proceedings of the Royal Society B: Biological Sciences, 276,21–30. We thank L. Nowak and D. Hanz for providing additional measure- Gill, F., & Donsker, D. (2015). IOC World Bird List (v 5.1). Retrieved from ment data and P. Grzeszkowiak for assisting with configuring supple- www.worldbirdnames.org/DOI-5/master_ioc_list_v5.1.xls mentary data tables. H. van Grouw and M. Adams (NHM Tring), K. Graham, C. H., & Fine, P. V. A. (2008). Phylogenetic beta diversity: Link- L. Smith Date (NMVM, Melbourne) and J. Fjeldsa (ZMUC Copen- ing ecological and evolutionary processes across space in time. Ecol- ogy Letters, 11, 1265–1277. hagen) provided access to bird collections kept in their charge. Our Graham, C. H., Ron, S. R., Santos, J. C., Schneider, C. J., & Moritz, C. work was supported by the German Research Foundation Emmy (2004). Integrating phylogenetics and environmental niche models to Noether Fellowship DFG FR 3246/2-1. explore speciation mechanisms in Dendrobatid frogs. Evolution, 58, 1781–1793. Grant, P. R., & Grant, B. R. (2006). Evolution of character displacement in ORCID Darwin’s finches. Science, 313, 224–226. Harmon, L. J., Kolbe, J. J., Cheverud, J. M., & Losos, J. B. (2005). Conver- Anna G. Phillips http://orcid.org/0000-0002-0183-4133 gence and the multidimensional niche. Evolution, 59, 409–421. Harmon, L. J., Losos, J. B., Davies, T. J., Gillespie, R. G., Gittleman, J. L., Jennings, W. B., ... Mooers, A. O. (2010). Early bursts of body size and shape evolution are rare in comparative data. Evolution, 64, REFERENCES 2385–2396. Ackerly, D. D. (2009). Conservatism and diversification of plant functional Harmon, L. J., Melville, J., Larson, A., & Losos, J. B. (2008). The role of traits: Evolutionary rates versus phylgoenetic signal. Proceedings of geography and ecological opportunity in the diversification of gay – the National Academy of Sciences USA, 106(Suppl), 19699–19706. geckos (Phelsuma). Systematic Biology, 57, 562 573. Barrera-Guzman, A. O., Mila, B., Sanchez-Gonzalez, L. A., & Navarro- Harmon, L. J., Schulte, J. A., Larson, A., & Losos, J. B. (2003). Tempo and – Siguenza,€ A. G. (2012). Speciation in an avian complex endemic to mode of evolutionary radiation in iguanian lizards. Science, 301, 961 the mountains of Middle America (Ergaticus, Aves: Parulidae). Molecu- 964. lar Phylogenetics and Evolution, 62, 907–920. Hawkins, B. A., Diniz-Filho, J. A. F., Jaramillo, C. A., & Soeller, S. A. Blomberg, S. P., Garland, T. Jr, & Ives, A. R. (2003). Testing for phyloge- (2007). Climate, niche conservatism, and the global bird diversity gra- – netic signal in comparative data: Behavioural traits are more labile. dient. The American Naturalist, 170(Suppl), S16 S27. Evolution, 57, 717–745. Holt, B. G., Lessard, J.-P., Borregaard, M. K., Fritz, S. A., Araujo, M. B., ... ’ Bohning-Gaese,€ K., & Oberrath, R. (1999). Phylogenetic effects on mor- Dimitrov, D., Rahbek, C. (2013). An update of Wallace s zoogeo- – phological, life-history, behavioural and ecological traits of birds. Evo- graphic regions of the world. Science, 339,74 79. lutionary Ecology Research, 1, 347–364. Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K., & Mooers, A. O. (2012). – Bravo, G. A., Remsen, J. V., & Brumfield, R. T. (2014). Adaptive processes The global diversity of birds in space and time. Nature, 491, 444 drive ecomorphological convergent evolution in antwrens 448. (Thamnophilidae). Evolution, 68, 2757–2774. Jønsson, K. A., Fabre, P.-H., Fritz, S. A., Etienne, R. S., Ricklefs, R. E., Calmaestra, R. G., & Moreno, E. (2000). Ecomorphological patterns Jørgensen, T. B., & Irestedt, M. (2012). Ecological and evolutionary related to migration: A comparative osteological study with passeri- determinants for the adaptive radiation of the Madagascan vangas. – nes. Journal of Zoology, 252, 495–501. Proceedings of the National Academy of Sciences USA, 109, 6620 Cardillo, M., & Warren, D. L. (2016). Analysing patterns of spatial and 6625. niche overlap among species at multiple resolutions. Global Ecology Jønsson, K. A., Lessard, J.-P., & Ricklefs, R. E. (2015). The evolution of and Biogeography, 25, 951–963. morphological diversity in continental assemblages of passerine birds. – Davies, T. J., Meiri, S., Barraclough, T. G., & Gittleman, J. L. (2007). Spe- Evolution, 69, 879 889. cies co-existence and character divergence across carnivores. Ecology Karr, J. R., & James, F. C. (1975). Eco-morphological configurations and Lettres, 10, 146–152. convergent evolution in species and communities. In M. Cody & J. – Dawideit, B. A., Phillimore, A. B., Laube, I., Leisler, B., & Bohning-Gaese,€ Diamond (Eds.), Ecology and evolution of communities (pp. 258 291). K. (2009). Ecomorphological predictors of natal dispersal distances in Cambridge, MA: Harvard University Press. birds. Journal of Ecology, 78, 388–395. Laliberte, E., & Legendre, P. (2010). A distance-based framework for mea- Dayan, T., & Simberloff, D. (2005). Ecological and community-wide char- suring functional diversity from multiple traits: A distance-based acter displacement: The next generation. Ecology Letters, 8, 875–894. framework for measuring from multiple traits functional diversity. – Dehling, D. M., Fritz, S. A., Topfer,€ T., P€ackert, M., Estler, P., Bohning-€ Ecology, 91, 299 305. € Gaese, K., & Schleuning, M. (2014). Functional and phylogenetic Laube, I., Graham, C., & Bohning-Gaese, K. (2013). Intra-generic spe- diversity and assemblage structure of frugivorous birds along an ele- cies richness and dispersal ability interact to determine geo- vational gradient in the tropical Andes. Ecography, 37, 1047–1055. graphic ranges of birds. Global Ecology and Biogeography, 22, – Drummond, A. J., Suchard, M. A., Xie, D., & Rambaut, A. (2012). Bayesian 223 232. phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Losos, J. B. (2008). Phylogenetic niche conservatism, phylogenetic signal Evolution, 29, 1969–1973. and the relationship between phylogenetic relatedness and ecological – Eck, S., Fiebig, J., Fiedler, W., Heynen, I., Nicolai, B., Topfer,€ T., ... Woog, similarity among species. Ecology Letters, 11, 995 1003. F. (2011). Measuring birds. Vogel€ vermessen. Deutsche Ornithologen- Lovette, I. J., Bermingham, E., & Ricklefs, R. E. (2002). Clade-specific mor- Gesellschaft. phological diversification and adaptive radiation in Hawaiian PHILLIPS ET AL. | 2347

songbirds. Proceedings of the Royal Society B: Biological Sciences, 269, Velasco, J. A., Martinez-Meyer, E., Flores-Villela, O., Garcıa, A., Algar, A. 37–42. C., Kohler,€ G., & Daza, J. M. (2016). Climatic niche attributes and MacArthur, R. (1958). Population ecology of some warblers of northeast- diversification in Anolis lizards. Journal of Biogeography, 43, 134–144. ern coniferous forests. Ecology, 39, 599–619. Verbruggen, H., Tyberghein, L., Pauly, K., Vlaeminck, C., Van Nieuwen- Maglianesi, M. A., Bluthgen,€ N., Bohning-Gaese,€ K., & Schleuning, M. huyze, K., Kooistra, W. H. C. F., ... de Clerck, O. (2009). Macroecol- (2014). Morphological traits determine specialization and resource ogy meets macroevolution: Evolutionary niche dynamics in the use in plant – Hummingbird networks in the neotropics. Ecology, 95, seaweed Halimeda. Global Ecology and Biogeography, 18, 393–405. 3325–3334. Warren, D. L., Cardillo, M., Rosauer, D. F., & Bolnick, D. I. (2014). Mistak- Mahler, D. L., Revell, L. J., Glor, R. E., & Losos, J. B. (2010). Ecological ing geography for biology: Inferring processes from species distribu- opportunity and the rate of morphological evolution in the diversifi- tions. Trends in Ecology & Evolution, 29, 572–580. cation of greater Antillean anoles. Evolution, 64, 2731–2745. Webb, C. O., Ackerly, D. D., McPeek, M. A., & Donoghue, M. J. (2002). Maire, E., Grenouillet, G., Brosse, S., & Villeger, S. (2015). How many Phylogenies and community ecology. Annual Review of Ecology and dimensions are needed to accurately assess functional diversity? A Systematics, 33, 475–505. pragmatic approach for assessing the quality of functional spaces. Wiens, J. J. (2011). The causes of species richness patterns across space, Global Ecology and Biogeography, 24, 728–740. time, and clades and the role of “ecological limits”. The Quarterly Miller, E. T., Zanne, A. E., & Ricklefs, R. E. (2013). Niche conservatism Review of Biology, 86,75–96. constrains Australian honeyeater assemblages in stressful environ- Williams, E. E. (1995). The origin of faunas. Evolution of lizard congeners ments. Ecology Letters, 16, 1186–1194. in a complex island fauna: A trial analysis. In T. Dobzhansky, M. K. Mittelbach, G. G., & Schemske, D. W. (2015). Ecological and evolutionary Hecht, & W. C. Steere (Eds.), Evolutionary biology (pp. 47–89). New perspectives on community assembly. Trends in Ecology & Evolution, York, NY: Springer US. 30, 241–247. Wilman, H., Belmaker, J., Simpson, J., de la Rosa, C., Rivadeneira, M. M., Near, T. J., & Benard, M. F. (2004). Rapid allopatric speciation in logperch & Jetz, W. (2014). EltonTraits 1.0: Species-level foraging attributes of darters (Percidae: Percina). Evolution, 58, 2798–2808. the world’s birds and mammals. Evolution, 95, 2027. Nyari, A. S., & Reddy, S. (2013). Comparative phyloclimatic analysis and Winkler, H., & Leisler, B. (1992). On the ecomorphology of migrants. Ibis, evolution of ecological niches in the scimitar babblers (Aves: Timali- 134,21–28. idae: Pomatorhinus). PLoS ONE, 8, e55629. Woodward, B. L., Winn, J. P., & Fish, F. E. (2006). Morphological special- Pagel, M. (1999). Inferring the historical patterns of biological evolution. izations of baleen whales associated with hydrodynamic performance Nature, 401, 877–884. and ecological niche. Journal of Morphology, 267, 1284–1294. Pigot, A. L., & Tobias, J. A. (2013). Species interactions constrain geo- graphic range expansion over evolutionary time. Ecology Letters, 16, 330–338. Price, T. (2008). Speciation in birds. Greenwood Village: Roberts and Com- BIOSKETCH pany. Revell, L. (2012). phytools: An R package for phylogenetic comparative Anna Phillips is a PhD student interested in vertebrate phylo- biology (and other things). Methods in Ecology and Evolution, 3, 217– geography and macroevolution. Her current work focuses on 223. inferring ecological and evolutionary patterns and processes from Ricklefs, R. E. (2004). Cladogenesis and morphological diversification in trait distributions of avian species. passerine birds. Nature, 430, 338–341. Ricklefs, R. E. (2012). Species richness and morphological diversity of Author contributions: A.G.P., S.F., T.T. and K.B.-G. designed the passerine birds. Proceedings of the National Academy of Sciences USA, 109, 14482–14487. study. A.G.P. collected and analysed the data, and led the writ- Ricklefs, R. E., & Miles, D. B. (1994). Ecological and evolutionary infer- ing. C.R. provided geographic data. All authors discussed the ences from morphology: An ecological perspective. In P. C. Wain- results and contributed to the writing. wright, & S. M. Reilly (Eds.), Ecological morphology (pp. 13–41). Chicago, IL: The University of Chicago Press. Rodrıguez-Girones, M. A., & Santamarıa, L. (2007). Resource competition, character displacement, and the evolution of deep corolla tubes. The American Naturalist, 170, 455–464. SUPPORTING INFORMATION Sidlauskas, B. (2008). Continuous and arrested morphological diversifica- Additional supporting information may be found online in the tion in sister clades of characiform fishes: A phylomorphospace approach. Evolution, 62, 3135–3156. Supporting Information section at the end of the article. Simoes,~ M., Breitkreuz, L., Alvarado, M., Baca, S., Cooper, J. C., Heins, L., ... Lieberman, B. S. (2016). The evolving theory of evolutionary radi- ations. Trends in Ecology & Evolution, 31,27–34. € Thorpe, R. S., Surget-Groba, Y., & Johansson, H. (2008). The relative How to cite this article: Phillips AG, Topfer T, Rahbek C, importance of ecology and geographic isolation for speciation in Bohning-Gaese€ K, Fritz SA. Effects of phylogeny and – anoles. Philosophical Transactions of the Royal Society B, 363, 3071 geography on ecomorphological traits in passerine bird clades. 3081. J Biogeogr. 2018;45:2337–2347. https://doi.org/10.1111/ Tobias, J. A., Cornwallis, C. K., Derryberry, E. P., Claramunt, S., Brumfield, R. T., & Seddon, N. (2014). Species coexistence and the dynamics of jbi.13383 phenotypic evolution in adaptive radiation. Nature, 506, 359–363.